The study, published in Frontiers in Network Physiology, set out to test a deceptively simple hypothesis. If ageing degrades the corticospinal pathway, and if stride-to-stride dynamics reflect the quality of motor control, then measures of corticospinal function should predict the fractal structure of walking variability in older adults. To find out, a team of researchers from Portugal and the United Kingdom recruited thirty-four older adults, seventeen men and seventeen women, with an average age of nearly seventy. All participants were free of neurological or neuromuscular disorders and were carefully screened for contraindications to brain stimulation. Each volunteer visited the laboratory for a single session lasting about an hour, during which the researchers probed the neural circuitry of the ankle with remarkable precision.

The technique at the heart of the experiment was transcranial magnetic stimulation, or TMS, a non-invasive method that uses rapidly changing magnetic fields to induce small electrical currents in the cerebral cortex. By positioning a double-cone coil over the leg area of the primary motor cortex, the researchers stimulated the neurons that control the tibialis anterior, the shin muscle that lifts the foot during the swing phase of walking. This muscle was chosen deliberately: it is primarily active during the phase of gait when the foot must clear the ground, and with ageing it becomes weaker and stiffer, reducing toe clearance and increasing the risk of tripping. The team quantified three distinct neurophysiological properties of this muscle’s motor pathway.

The first measure was corticospinal excitability, indexed by the amplitude of the motor evoked potential, the electrical burst recorded from the muscle when a pulse of magnetic stimulation triggers a volley of activity descending from the cortex to the spinal cord. The second was the duration of the cortical silent period, the temporary pause in voluntary muscle activity that follows a stimulus, which reflects GABA-B-mediated inhibitory processes within the cortex. The third was short-interval intracortical inhibition, measured with a paired-pulse protocol in which a subthreshold conditioning stimulus suppresses the response to a subsequent suprathreshold test stimulus delivered three milliseconds later, providing a window onto GABA-A-mediated inhibitory interneuron circuits. Together, these three measures capture the excitatory and inhibitory architecture of the corticospinal system with a resolution that few other techniques can match.

Immediately after the stimulation session, participants walked on an instrumented treadmill for twelve minutes at their preferred speed, determined individually through a careful speed-adjustment procedure. Ground reaction forces measured by the treadmill’s force plates allowed the researchers to detect heel strikes with millisecond precision, using an eight-to-twelve-newton force threshold. From each recording they excluded the first thirty seconds to avoid the transient adjustment period, then extracted roughly six hundred consecutive inter-stride intervals per person, a sample size consistent with recommendations for reliable nonlinear analysis. Each of these time series was then subjected to detrended fluctuation analysis, a modified random-walk method originally developed for heartbeat rhythms.

Detrended fluctuation analysis works by integrating the stride-interval series, dividing it into windows of varying length, fitting and removing the local trend within each window, and then quantifying the root-mean-square fluctuations across window sizes. The slope of the relationship between fluctuation magnitude and window size, plotted on logarithmic axes, yields the fractal scaling exponent alpha. An alpha value greater than 0.5 indicates persistent long-range correlations, meaning that increases in stride interval tend to follow increases and decreases follow decreases, a signature of adaptive, flexible control. Values below 0.5 suggest anti-persistent, tightly regulated dynamics, while a value near 0.5 approximates uncorrelated randomness. The older adults in this study produced an average alpha of 0.79, confirming that their gait dynamics had indeed drifted below the threshold of roughly 0.86 that previous research has proposed to distinguish healthy young from older walkers.

Then came the surprise. When the researchers fed the three corticospinal measures into a multiple linear regression model to predict the fractal exponent, the model collapsed. It explained just over two percent of the variance in gait dynamics, a figure that was statistically indistinguishable from zero. Corticospinal excitability, silent period duration, and short-interval intracortical inhibition were all non-significant predictors, each with a p-value far above the conventional 0.05 threshold. The authors had hypothesised the opposite, expecting that age-related neural decrements in the corticospinal tract would be a determinant of gait’s motor control. The data refused to cooperate.

Why might the brain’s flagship motor pathway fail to explain the rhythm of ageing steps? The researchers point to the architecture of locomotor control itself. Walking is fundamentally orchestrated by spinal central pattern generators, networks of interneurons that produce the core rhythmic alternation of flexion and extension without any conscious input. These spinal circuits are modulated by afferent sensory feedback from proprioceptors and skin receptors and by descending signals from the brainstem and cortex. The basal ganglia, cerebellum, and brainstem cooperate to regulate muscle tone and rhythmic limb activity automatically. In this distributed scheme, the corticospinal tract may simply not be the dominant player during steady-state, highly automated walking. The authors suggest that the reticulospinal tract, a bilateral brainstem pathway that projects broadly to large muscle groups and governs postural tone, may contribute far more to routine gait than traditionally assumed, a possibility supported by studies in spinal cord injury showing compensatory increases in reticulospinal excitability. Corticospinal drive, by contrast, appears to matter most when walking demands conscious adaptation, such as split-belt treadmill walking or visually guided navigation.

The findings also resonate with a broader shift in how scientists conceptualise ageing and movement. Evidence now links age-related gait changes not just to any single pathway but to the integrity of distributed brain networks, including grey matter regions supporting memory and executive function, and even to the white-matter architecture of inhibitory control circuits connected to the subthalamic nucleus. Meanwhile, peripheral factors such as sarcopenia, sensory nerve loss, joint stiffness, motor neuron loss, and the progressive decline of motor units all reshape the neuromuscular periphery. Fractal gait dynamics, the authors argue, likely emerge from the interaction of all of these levels at once, and no single snapshot of corticospinal function, taken while a participant sits still performing an isometric ankle contraction rather than actually walking, can capture that emergent complexity.

The study has honest limitations that its authors acknowledge. The neuromuscular assessment was performed during a static contraction, and corticospinal measures are strongly task-dependent, so they may not fully represent corticospinal activity during dynamic locomotion. Measuring TMS during actual walking is technically fraught, since the head’s vertical displacement disrupts coil stability and the rhythmic clicking of stimulation can itself entrain gait. Treadmill walking, even at preferred speed, is not identical to overground walking, the sample was modest, and no covariates were adjusted. Yet the null result is precisely what makes the study valuable. It redirects attention away from a single descending pathway and toward the multilevel network of spinal, supraspinal, and peripheral mechanisms that collectively generate the fractal texture of human gait. For the millions of older adults at risk of falls, that shift in perspective could ultimately guide rehabilitation strategies that train the whole adaptive system rather than one neural cable.

Subject of Research: Corticospinal function and stride-to-stride gait dynamics in older adults

Article Title: Corticospinal function does not predict stride-to-stride gait dynamics in older adults

Article References: Gomes, J. S., Oliveira, J. H., Bauer, P., Sousa, L., Sel, A., Cortes, N., & Vaz, J. R. (2026). Corticospinal function does not predict stride-to-stride gait dynamics in older adults. Frontiers in Network Physiology, 6, Article 1946459. https://doi.org/10.3389/fnetp.2026.1946459

Image Credits: AI Generated

DOI: 10.3389/fnetp.2026.1946459

Keywords: ageing, gait, corticospinal tract, transcranial magnetic stimulation, stride-to-stride fluctuations, fractal scaling, motor control, fall risk, central pattern generators, reticulospinal tract, movement variability, network physiology